Claire's CTO: 90-Day Priorities
Days 1-30 · Freeze Salesforce, design the new spine
One job: make every irreversible architecture decision before any spend or migration begins, and stop the bleeding of new investment into a stack that has to be removed.
Week 1 Authority, audit, alignment
- Day 1 all-hands directive: "Effective today, we are systematically removing Salesforce. No new Salesforce objects, automations, or integrations are being created. All new initiatives target HubSpot or platform-neutral components." Tone is decisive, not punitive. This is strategy, not blame.
- Days 2-5, Salesforce audit. Inventory every standard and custom object, every workflow, every automation, every integration, every active user. Tag each artifact Critical, Useful, or Junk.
- Days 5-7, stakeholder interviews. Sit with finance, supply chain, merchandising, store operations, and e-commerce. Document what they need, what they hate about Salesforce, and what data only lives in Salesforce today. Those are the migration risks.
Week 2 Lock technology decisions
- Days 8-10, lock the platform stack. CRM and revenue brain: HubSpot. E-commerce front end: Shopify or BigCommerce. POS: Lightspeed Retail (or equivalent cloud POS with a native or proven HubSpot connector). Data warehouse: Azure.
- Days 10-12, draft and approve the one-page target architecture. HubSpot at the center; e-commerce, POS, OMS feeding into it; Azure receiving every transaction; Power BI dashboards and a decision-making AI layer on top.
- Days 12-14, timeline and resourcing. 18-month roadmap with Phase 0 through Phase 3, named team requirements, $105-165M total investment envelope.
Weeks 3-4 Design and specification
- Days 15-21, HubSpot data model. Hubs in scope (Sales, Marketing, Service, Operations), deal pipelines by channel and region, lifecycle stages (tween, teen, young adult, parent; self-purchase, gifting, piercing), custom properties for segmentation and family-relationship modeling, permission model. Simplify, do not mirror Salesforce's complexity.
- Days 21-25, data mapping workbook. Field-by-field mapping from Salesforce to HubSpot. Explicit list of what will not be migrated. Migration order: companies, contacts, deals, tickets, activity history.
- Days 25-28, dashboard specification. Minimum viable metrics: revenue, units, gross margin by channel, region, product category, segment, refreshed daily or near real-time. Store-manager view, merchant view, finance view. One page.
Days 31-60 · Stand up the spine, begin migration
Phase 1 is execution. HubSpot moves into production. E-commerce flows in. POS flows in. Azure receives everything. The first dashboard goes live. Salesforce migration is piloted on a controlled slice. By Day 60, the new spine is real.
Weeks 5-6 HubSpot config + e-commerce integration
- Days 31-35, HubSpot production setup. Production environment, users and roles, pipelines, lifecycle stages, custom properties and objects. Validation gate: can a deal be created, moved through pipeline, and reported on cleanly?
- Days 35-40, e-commerce integration. Choose connector (native HubSpot, Unific, or partner). Configure sync for customers, products, orders, abandoned carts. Validation: new online orders appear in HubSpot within 15 minutes; clean, no duplicates.
Weeks 7-8 POS, warehouse, first dashboard
- Days 41-45, POS deployment. Lightspeed Retail (or equivalent) in pilot stores. Configure transaction sync: store, cashier, line items, customer, timestamp. In-store receipts in HubSpot within 15-30 minutes.
- Days 45-50, Azure warehouse. Schema covering e-commerce and POS in one combined transaction table. Pipelines on hourly or 4-hour micro-batch. Every transaction in the warehouse within 4 hours.
- Days 50-55, dashboard implementation. Power BI on the warehouse schema. Daily revenue, units, gross margin by channel, region, category. Store and merchant dashboards. Training to store ops, merchants, finance.
Parallel track Salesforce migration planning + pilot
- Days 31-40, data preparation. Export Salesforce data, identify quality issues (duplicates, inconsistent formats, orphan records), clean (standardize picklists, dedupe, fix emails), disable any Salesforce automations that modify records during export.
- Days 41-50, pilot migration. One region (e.g., Midwest with 3-4 territories) or 5-10% of accounts. Migrate companies, then contacts, then deals into HubSpot. Reconcile. Capture and fix mapping bugs.
- Days 50-60, full migration prep. Sign off the playbook with team leads. Phase the full migration. Secure leadership commitment that all new work goes into HubSpot starting Day 61.
Days 61-90 · Cut over Salesforce, launch decision AI
Two things in parallel: kill Salesforce, and prove decision-making AI works on a real category. Each is independently board-relevant; together they reframe what Claire's technology is.
Weeks 9-10 Salesforce cutover
- Days 61-65, full migration execution. Run the validated playbook: companies, contacts, deals, tickets in phased waves. Set Salesforce to read-only. No more changes in Salesforce, period.
- Days 65-70, cutover and stabilization. All new deals, contacts, activities go into HubSpot only. Train users on HubSpot workflows. Daily reconciliation. Fix operational breakdowns immediately.
- Days 70-80, parallel run. Salesforce stays read-only for 30 days. All operational work in HubSpot. Reconciliation reports twice weekly.
- Days 81-90, decommission planning. After 30 clean days, plan the Salesforce shutdown. Archive for compliance. Begin canceling licenses (recurring savings of $10-15M annually). Shut down Salesforce-to-ERP integrations.
Weeks 10-12 Decision-making AI pilot
Pilot deliberately narrow: earrings as the category (35%+ of jewelry revenue, high velocity, clear seasonality), Midwest as the region (200-300 stores, mixed urban/suburban/mall), teen self-purchase as the segment. Six-step loop: forecast, optimize, propose, human review, execute, learn.
- Days 61-75, model development in parallel with Salesforce cutover. 24 months of earring transactions, current inventory, vendor lead times, MOQs, margin and seasonality data from Azure. Demand forecast model (time-series with promotional, price-elasticity, seasonality inputs). Optimization routine (linear programming over budget, safety stock, MOQs, DC capacity).
- Days 76-85, deployment. Ranked, explained recommendations in the OMS UI. Each carries quantity, expected sell-through, overstock risk, margin impact, and rationale. Merchants see them daily.
- Days 85-90, human-in-loop review. Merchants approve, modify, or reject. Every override carries a captured reason. Those reasons feed model retraining. Measurement: forecast accuracy (target below 15% error against 25-30% baseline), sell-through (+5-10%), overstock (-20%), margin (+1-2 points).
Day 90 Board Proof Points
The Day 90 board pack is short and concrete. Four claims, each with evidence.
Salesforce is gone
HubSpot is the system of record for customers, deals, and revenue. Annual licensing run-rate reduced by $10-15M. Migration executed under a validated playbook with daily reconciliation; no operational breakdowns survived past 48 hours.
Omnichannel data is live
Every e-commerce and POS transaction lands in the Azure warehouse within four hours and surfaces in dashboards within 24. Finance, merchandising, and store operations all read from the same source. There is no longer a debate about what "revenue" means.
Decision-making AI is generating real buying decisions
For earrings in the Midwest, the system produces ranked, explained purchase order recommendations weekly. Merchants approve, modify, or reject with reasons captured. By Month 6, scales to necklaces and a second region. By Month 12, 60% of assortment.
The 18-month plan is on track
Phase 0 complete. Phase 1 in execution. The path to 25%+ e-commerce penetration, 300+ remodeled stores, EBITDA break-even by Month 18 is defended by operating evidence, not promises. $105-165M investment recovering against $500-800M annualized revenue at 60-70% probability.
Named Decisions Locked in 90 Days
The 90 days are organized around a small number of irreversible decisions made early. Locking them is what allows everything downstream to move at the speed the plan requires.
- CRM and revenue brain (Week 2): HubSpot, not Salesforce. TCO 50-60% lower at retail scale. Native integrations into chosen e-commerce and POS partners.
- E-commerce platform (Week 2): Shopify or BigCommerce. Both have proven HubSpot integrations and migrate faster than Salesforce Commerce Cloud allows.
- POS: Lightspeed Retail or equivalent cloud POS with a real HubSpot connector. Pilot stores by Day 60, rolling out from there.
- Data warehouse: Azure, consistent with existing cost-optimization work and current operating capability.
- Decision-making AI category and region: Earrings in the Midwest, teen self-purchase segment. Velocity, seasonality, and signal clarity make the pilot defensible.
Each decision is a single sentence in the Day 30 board pack with reasoning documented and alternatives explicitly considered. The point of locking them early is to remove the litigation that otherwise consumes Phase 1.
Risk Register and Mitigation
Stakeholder Management Cadence
What Day 90 Sets Up
After Day 90, Phase 1 continues for another three months. POS rolls to 300+ stores. Decision-making AI extends to necklaces and a second region. The first 20-30 Gen Z store remodels go live in priority markets. The Generations program launches in 5-10 pilot stores, with the family-relationship CRM model captured and the first tea-house partnership signed. The Princesses brand expression, the first-piercing ritual, moves from pilot to standard. By Month 6 the e-commerce penetration target shifts from 12% to 15%, same-store sales decline narrows from -15% to -8%, and inventory turns improve 5%.
The 90 days are not the transformation. They are the proof that the transformation is real, repeatable, and on schedule. That proof is what unlocks the rest of the capital and the rest of the timeline.